This proposal describes our plans to build a set of software tools and an online capability for generating high-resolution, study-specific atlases from the ADNI database of images and patient data. The proposed tools will rely on the technology for large deformation diffeomorphic metric mappings (LDDMMs), to generate high-quality, multimodal brain atlases. The goal of this project is not merely a single atlas nor is it simply a set of new algorithms. Rather, we will provide an online service through which users can request custom atlases that they specify through a web interface. A computational infrastructure will deliver these atlases to the user. We will also pursue fundamental and technological improvements in the atlas construction framework as well as the validation of these technologies. In order to provide a clinical context for these developments, we will use these tools to conduct an atlas-based study on metabolic asymmetry in dementia with coinvestigators in the Department of Neurology at the University of Utah. PUBLIC HEALTH RELEVANCE: The goals of this project are multifold. We will be developing open-source software for generating LDDMMs and a web interface open to researchers around the world for automatically generating and delivering high-quality, on-demand atlases from ADNI database queries. The proposed Brain Atlas Portal will produce, as specified by the user: (i) multimodal imagery, with transformations between subjects and the mean, (ii) reliability index for the atlas, (iii) quantification of structural and metabolic variability of the cohort, and (iv) a set of longitudinal multimodal atlases for a cohort.